When Zeus, a medical student in central Nigeria, returns home from hospital shifts, he transforms his studio apartment into a makeshift motion-capture studio. He straps his iPhone to his forehead, activates a ring light, and records himself performing mundane actions—walking, climbing stairs, reaching, gesturing—uploading the videos to earn between $5 and $15 per submission. Zeus is one of thousands of gig workers globally hired by robotics companies to generate training data for humanoid robots, a labor arrangement that has largely escaped regulatory scrutiny. The workers, scattered across Nigeria, Kenya, the Philippines, and other low-wage nations, are classified as independent contractors rather than employees, a designation that sidesteps payroll taxes, benefits, and labor protections that would otherwise apply under U.S. or European employment law.

This model mirrors earlier patterns in AI training—companies like OpenAI relied on outsourced contractors to label data for large language models—but humanoid training introduces new ethical dimensions. Unlike static image labeling, video data collection requires workers to perform repetitive physical motions for extended periods, raising ergonomic and health concerns that remain unaddressed in most contracts. Additionally, workers typically retain no ownership over the biometric data they generate, and most sign non-disclosure agreements preventing them from discussing compensation or working conditions. A 2023 investigation by researcher found that companies deploying this strategy rarely conduct labor audits or verify that subcontractors comply with local wage minimums. The U.S. Fair Labor Standards Act (FLSA) contains a broad independent-contractor exemption for certain knowledge work, but courts have increasingly questioned whether gig workers misclassified as contractors should receive employee protections. No robotics firm has yet faced enforcement action under this standard, partly because the labor department has historically focused on domestic gig platforms like Uber rather than international data collection networks.

Policy momentum is beginning to shift. The EU AI Act, which took effect in 2024, mandates transparency about training datasets and human oversight for high-risk systems, potentially requiring robotics firms to disclose labor practices. Simultaneously, shareholder advocacy groups have begun pressuring companies like Boston Dynamics to publish labor audits for their training contractors. The critical gap remains: no international labor standard currently governs remote gig work in AI and robotics training. A federal labor department investigation into humanoid-training platforms' contractor classifications could establish precedent, as could collective action from worker advocacy groups in Nigeria and the Philippines. Until regulators act, thousands of workers will continue training the robots of tomorrow while remaining invisible to the protections designed to shield them.